Speaker, Prof. Mark Sanderson, RMIT University, Australia
Abstract. This year, (2013) marks the 20th anniversary of the
first public web search engine JumpStation launched in late
1993. For those who were around in those early days, it was
becoming clear that an information provision and an information
access revolution was on its way; though very few, if any would
have predicted the state of the information society we have
today. It is perhaps worth reflecting on what has been achieved
in the field of information retrieval since these systems were
first created, and consider what remains to be accomplished. It
is perhaps easy to see the success of systems like Google and
ask what else is there to achieve? However, in some ways, Google
has it easy. In this talk, I will explain why Web search can be
viewed as a relatively easy task and why other forms of search
are much harder to perform accurately.
Search engines require a great deal of tuning, currently
achieved empirically. The tuning carried out depends greatly on
the types of queries submitted to a search engine and the types
of document collections the queries will search over. It should
be possible to study the population of queries and documents and
predictively configure a search engine. However, there is little
understanding in either the research or practitioner communities
on how query and collection properties map to search engine
configurations. I will present the some of the early work we
have conducted at RMIT to start charting the problems in this
particular space.
Another crucial challenge for search engine companies is how
to ensure that users are delivered the best quality content.
There is a growth in systems that recommend content based not
only on queries, but also on user context. The problem is that
the quality of these systems is highly variable; one way of
tackling this problem is gathering context from a wider range of
places. I will present some of the possible new approaches to
providing that context to search engines. Here diverse social
media, and advances in location technologies will be emphasized.
Finally, I will describe what I see as one of the more
important challenges that face the whole of the information
community, namely the penetration of computer systems to
virtually every person on the planet and the challenges that
such an expansion presents.
About the Speaker:
Mark Sanderson is Professor at RMIT University in the School of
Computer Science and Information Technology. He is a researcher in
information retrieval (IR) (e.g. web search engines), and
particularly interested in evaluation of search engines, but also
work in geographic search, cross language IR (CLIR),
summarisation, image retrieval by captions, word sense ambiguity.
Prof. Sanderson has a number of active research projects and a
number of research students Editorial, including - Associate
Editor of ACM Transactions on the Web and Information Processing
and Management; co-editor of Foundations and Trends in
Information Retrieval; the co-PC Chair of ACM SIGIR 2012. Prof.
Sanderson is visiting professor at NII.
http://www.seg.rmit.edu.au/mark/
Academic Host: Noriko Kando, NII
Inquiry: kando-secr AT nii ac jp / Phone 03-4212-2733